Document Type

Article

Disciplines

Econometrics | Finance | Macroeconomics

Abstract

This study seeks to determine whether the expectation of a government bailout in the event of a crisis incentivizes ex ante risk-taking in banks. Although this subject has been the subject of extensive research in both economics and finance, this specific paper differs from existing literature in that it uses a strictly empirical methodology with minimal theoretical modeling and few assumptions about the banking sector, while also considering the entire banking sector and limiting itself exclusively to ex ante bank investment decisions. This study employs data compiled from various sources about 11 American banks of various sizes between January 2002 and September 2008. Based on the results of Bayazitova and Shivdasani (2012), this study uses six variables as proxies for bailout probability and combines them into a single bailout probability composite index: Total assets, notional derivatives relative to assets, the wholesale debt ratio, the loan-loss provision ratio, write-downs relative to assets, and CEO compensation in excess of $500,000. It uses the stock beta to measure risk-taking. This study first performs a heteroscedasticity-robust OLS regression of the beta on the bailout probability index, then adds three control variables to that regression, and finally removes outliers from the dataset. The first two regressions yield a positive and statistically significant relationship between bailout probability and systematic risk, whereas the third yields a statistically insignificant one. As such, this study cannot demonstrate that bailout expectations influence banks’ investment decisions. Despite having several limitations, these results may have significant public policy implications and offer many opportunities for future research.

Download

Share

COinS